Non-linear Decision Feedback Equalizer for ISI Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
High-speed digital communication systems face challenges in correcting for dispersion and intersymbol interference (ISI) due to the time-varying nature of communication channels and the difficulty in designing analog-to-digital converters (ADCs) that operate at high baud rates, which affects the accuracy of timing recovery operations.
Innovation Solution
A signal processing system that includes an interleaved ADC and equalizer architecture, utilizing a baud rate phase detector and a time-varying phase detector to account for channel variations, and employing a decision feedback equalizer to correct for signal distortions introduced by the communication channel.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a serial ADC operates at high baud rates (excess of 1.5-2 GHz), then the transmission rate and bandwidth of the telecommunication system can be increased, but it becomes difficult and expensive to build the ADC
Solution Approach 1:
The patent divides the high-speed ADC operation into multiple parallel lower-speed ADCs operating in an interleaved manner. Each ADC operates at a manageable clock rate while collectively they achieve the required high baud rate sampling through time-interleaved operation, thus avoiding the complexity of building a single high-speed ADC
Solution Approach 2:
The patent transitions from a single-time-dimensional serial ADC operation to a multi-dimensional parallel structure where multiple ADCs operate simultaneously at different time offsets. This dimensional expansion allows the system to achieve high effective sampling rates without requiring any single ADC to operate at prohibitively high speeds
2Adaptability or versatility
If the channel characteristic is time-varying, then the communication system can adapt to changing conditions, but the timing recovery operations become less accurate
Solution Approach 1:
The patent performs preliminary channel characterization and equalization before the actual data transmission and timing recovery processes. By pre-adapting to the channel conditions and establishing initial equalizer coefficients, the system creates a more stable foundation for subsequent timing recovery operations, reducing the impact of time-varying channel effects
Solution Approach 2:
The patent implements feedback mechanisms where the equalizer continuously monitors the received signal quality and adapts its coefficients in real-time to compensate for channel variations. This feedback loop maintains timing recovery accuracy despite the time-varying nature of the communication channel by dynamically adjusting to changing conditions
3Reliability
If a decision feedback equalizer is used to correct for post-cursor ISI, then the signal distortion correction is improved, but the device complexity increases
Solution Approach 1:
The patent employs dynamic coefficient adaptation in the decision feedback equalizer where the feedback coefficients are continuously updated based on the detected signal decisions and error metrics. This dynamic adjustment allows the equalizer to maintain high correction accuracy for post-cursor ISI while optimizing the computational complexity by adapting only the necessary coefficients rather than maintaining a fixed complex structure
Data Source
AI summary
Embodiments include a decision feedback equalizer (DFE) that includes a first comparator configured to receive as inputs a soft value and a first threshold, a second comparator configured to receive as inputs the soft value and a second threshold, a selector configured to select an output of either the first comparator or the second comparator as a DFE output based on one or more previous bits output by the selector; an error calculator configured to determine an error for the first comparator and the second comparator, and a threshold adjuster configured to adjust the first threshold and the second threshold, the first threshold and the second threshold each being a non-linear combination of one or more previous outputs of the selector.